The Euler top and canonical lifts (Q1753820)

In this paper, the authors present an analogue for the multiplicative group \(\mathbb{G}_m\), of one of the main finiteness results for elliptic curves proved in [\textit{A. Buium} and \textit{B. Poonen}, Compos. Math. 145, No. 3, 566--602 (2009; Zbl 1232.11073)]. This is then applied to a question left open in [\textit{A. Buium} and \textit{E. Previato}, J. Number Theory 173, 37--63 (2017; Zbl 1358.14030), Remark 6.6], namely whether the main theorem, [loc. cit., Theorem 6.1] could be improved by showing that a certain extension property (saying, roughly, that an arithmetic flow on an affine elliptic fibration can be compactified) holds for the arithmetic Euler top, which was there introduced. As it is seen in this paper the answer to this question is negative: no such extension property holds and, hence, no such improvement of Theorem 6.1 in [loc. cit.] is possible. On the other hand the corresponding extension property does hold in the case of the classical Euler top, which is an algebraically completely integrable Hamiltonian system. In this paper, the authors prove a finiteness result for fibers that are canonical lifts in a given elliptic fibration. The question was motivated by the authors construction of an arithmetic Euler top, and it highlights an interesting discrepancy between the arithmetic and the classical case: in the former, it is impossible to extend the flows to a compactification of the phase space, viewed as an elliptic fibration over the space of action variables.